Gut, 1978, 20, 780-786
In vitro electrical activity in canine colon' N. L. SHEARIN, K. L. BOWES,2 AND Y. J. KINGMA From the Surgical-Medical Research Institute and Departments of Surgery and Electrical Engineering, University of Alberta, Edmonton, Alberta, Canada
In vitro slow wave activity was studied in strips of right and left canine colon with silver/silver chloride electrodes. Using visual and computer analysis, slow wave frequency and coupling was assessed between different recording sites and the effect of a cholinergic agonist on coupling and frequency was determined. A regular slow wave was always found to be present. Frequency in the left colon was slightly higher than in the right with a slight decline noted with time. Spike activity was rarely seen in unstimulated specimens. Administration of a cholinergic agonist produced a decrease in frequency with no improvement in coupling. Coupling was usually better in a circular than in a longitudinal direction. It was concluded that if electrical activity is important in the control of colon contractions, it is more likely to be involved in the control of segmentation than in propagated contractions. SUMMARY
Electrical activity in the colon has been studied in vivo and in vitro in different experimental animals and in humans (Couturier et al., 1969; Weinbeck et al., 1972; Taylor et al., 1975; Snape et al., 1976). In most of these studies electrical slow waves have been recognised only intermittently at two or more frequencies. However, Christensen et al. (1969) found electrical slow wave activity in the in vitro cat colon to be omnipresent at 005 Hz (3 c/m). We have studied the in vitro electrical activity of muscle strips from canine colon. Coupling between different recording sites was assessed to determine whether electrical activity could be responsible for propagated contractions. Methods
Strips of colon (15 x 8 cm) were removed from healthy dogs anaesthetised with chloralose. The mucosa was removed by sharp dissection and the tissue placed in one of two tissue chambers (15 x *5 x 13 cm). A Krebs-Ringer solution kept at 370-37-50 and aerated with a gas mixture of 95% 02 and 5 % CO2 was allowed to flow through the tissue chambers and over the tissue at a flow rate of 5-6 ml/m. Eight glass insulated silver/silver-chloride 'Supported by the Medical Research Council of Canada. 2Address for reprint requests: Kenneth L. Bowes, MD, Department of Surgery, 11-1 17 Clinical Sciences Bldg, University of Alberta, Edmonton, Alberta, Canada, T6G 2G3. Received for publication 6 April 1979
electrodes (1 mm diameter) were gently pressed on to the circular muscle tissue surface until a satisfactory record was obtained. The electrodes were separated by a distance of 1I0 cm. A heavy silver wire was placed in the tissue chamber to serve as a reference electrode. This apparatus is similar to that previously described by Christensen et al. (1969). Recordings of electrical activity were made on a Beckman Dynograph R 411 polygraph (Beckman Instruments Inc., Shiller Park, Illinois, USA) with filters set to produce a pass band from 0 2 (12 cpm) to 30 Hz. The signals were stored on an SR 300 Ampex tape recorder (Ampex, Redwood City, Calif., USA). The low cutoff frequency is the result of the time constant which is introduced when the recorder is used in the a.c. coupled mode. This mode of coupling was required because of the large d.c. offsets which tend to fluctuate. On the equipment available the a.c. coupling introduced a time constant of one secondthat is, a corner frequency of 0 16 Hz. The roll off towards frequencies less than 0 16 Hz is not more than 6 dB/octave. Signals with a period of 0 05 Hz, for example, are attenuated to about one-third of their original value. A total of 54 strips of colon were studied. Thirtyfour strips were analysed in detail; 11 were cut in a circular direction (five right colon; six left colon) and 23 in a longitudinal direction (14 right colon; nine left colon). A 30 minute basal record was obtained from each specimen. The effect of neostigmine methylsulphate (0 3 Fm/ml) in the perfusion fluid was then determined. In four paired colon specimens (right and
780
781
In vitro electrical activity i'n cani'ne colon 0 0
r-I
0r~~~~
0
CC,3) v 0
0OC, .
0 0
0
0$8
0
0(O .
0
LO
0 0
0 0
CO) 0 0
CMJ
0
0
c,0
a
000 oo 0
0 0 e--i
K
0
:0 000co ILnne-)XXXXHHHnu
Fig.
0*0 00 000000 00
Typical frequency
(0-Cane
08$
Channel ;L
0l08888880-6-12-0-Bdooooooooooob'db-dooooooooooooooooooooOO00000-
power spectra
2;1
with the maximum values normalised
of the first channel having that value
of electrical activity recorded
to 1000 units.
is
left colon from four dogs) basal
was
If the values
in vitro
9
J0.
in two
from three
cycles/mmn.
or more
sites
sites
cm
apart
Vertical axis: arbitrary are
units
identical, only the symbol
printed.
recordings
continued for six hours. Electrical slow quency
orizontal axis;
Canel3)
determined manually by
wave
were
fre-
counting three
results
expressed as the number of specimens in same frequency was present in the first three recording electrodes (frequency comparison). are
which the
non-consecutive five minute segments of record from
The distance between the two outer electrodes
each electrode. A five minute interval
2
was
chosen
during periods of this length the frequencies significantly and because the interval was large enough for beat to beat variations to be averaged out. The records were also analysed on a digital computer. Before processing the data, the taped analogue signals from the instrumentation recorder were lowpass ifiltered (0-5 Hz cut-off frequency), digitised (sampling rate of one per second) and stored on magnetic tape. The digitised data were then processed on the digital computer to obtain the frequency power spectra (Jenkins and Watts, 1968). Dominant frequencies were obtained from these frequency power spectra. A typical frequency spectrum is shown in Fig. 1. Coupling between different recording sites was assessed by comparing frequencies and phase and determining coherence and cross-correlation. The first two assessments were made visually; the latter two by computer. 1. Frequencies obtained from different recording electrodes on the same specimen were compared. The because
did not appear to vary
was
cm.
2. The number of shift between two
cycles during which recording electrodes 2
the
phase
cm
apart
remained within 3600. 3. Coherence between
trodes, 2 quency
cm
apart
bands
was
(0-033
signals from pairs of eleccalculated for
Hz)
on
the
narrow
computer.
freCo-
herence is
a measure of the linear dependence of one signal on another. If one signal is completely dependent on the other, coherence is unity. Other-
wise the coherence may have values between
zero
and
unity. The mathematical definition of coherence (which holds for stationary processes only) is given by:
Where:
G,,,(f)
Gyy(f)
and
trum of the
G,,y(f)= one-sided the two
one-sided measured power spec-
signals. measured cross-power spectrum of
signals.
Coherence
was
determined in the frequency bands in
782
~ ~ m=
\.1 V... IF
1
n
N. L. Shearin, K. L. Bowes, and Y. J. Kingma
which the power spectra of both signals were maximal. The cross spectral density function G,y(f) is a complex function and can be expressed as:
1
lil
1'
IC.Y...
Gxy(f) = Cxy(f) +jOxy(f)
Where:
j=v-I Cxy(f)=co-spectral density function. Oxy(f)=quadrature spectral density function.
2
3
Fig. 2 Cross-correlation between pairs of channels. Horizontal axis: timeshift, I per division. Vertical axis: Arbitrary units. 1: good correlation. 2: fair correlation. OI
I1 11ittillll'lliiiili
I
III jII
11 IIIIIIIjJIIM II lti
ltil
lil III] III
IllMIli L
11 li til ll
11I 111
III11
IIIltl 11 I111
One may also write:
Gxy(f)= Gxy(f) ejOxy(f)
Where:
Gxy(f) = [CXY2(f) + OXY2(f)]
and:
Oxy(f)=arc tan Oxy(f) CXY(f)
In the case under consideration, Oxy(f) represents the phase angle as a function of frequency between the two signals (Bendat and Piersol, 1966). 4. Cross-correlation of slow wave activity measured at electrodes spaced 2 cm apart was calculated on the digital computer. Cross-correlation as a function of the time shift with X was displayed on a large scieen cathode ray tube and then photographed. Observations of the cross-correlation graph on the CRT enable classifying the results into one of three categories of coupling: good, fair, poor. An example is shown in Fig. 2. To facilitate comparison of I~I iI
iII4,Il
IIIltl1 11 III 1III I: '11 Illlitllls 111 IMHIIii111 L
I'ilIll
,ll
IIIIIII
I. {~~~~~~~~~~~~~ 1 min
Fig. 3 Electrical slow wave activity of the type most commonly observed in recordings from canine colon. Electrodes are arranged in a longitudinal orientation on the colon strips (Ch. 1-4= left colon; Ch. 5-7-= right colon).
In vitro electrical activity in canine colo7
783
results, a numerical grading system was used: (good=100%; fair=50%; poor=0%). The period (and thus frequency) and the phase shift between channels could also be obtained from the graphs. The cross-correlation function is defined by: +T (t) rnim 1 J fi(t)f2 (t-r)dr T->oo 2T -T Where f,(t) and f2(t) are the signals to be correlated. 5. Phase shift between different recording sites
was determined visually by using the relation
9=t x 3600, where t is the delay between the onset T of slow wave depolarisation at different recording sites and T is the period of a complete wave cycle. The phase angle was calculated only when coupling was present. Phase shift was calculated on the computer in those specimens in which the coherence value exceeded 0 5.
Results
A regular electrical slow wave was always found in strips of canine colon muscle studied in vitro (Fig. 3). Frequency in the left colon was only slightly higher than that in the right colon in both paired (4 90 ±0-63 and 4-13 +0-30 c/m) and unpaired (5-02 0-35 and 4-81 ±0-38 c/m) experiments. The differences were not significant (P >0.1). These frequencies are only slightly lower than those noted in the chronic in vivo dog colon (Bowes et al., 1978). The in vitro frequency was well maintained in the tissue chamber. Only a slight decline was noted with time (Fig. 4). Spike activity was rarely seen in unstimulated specimens. Although slow waves were always recognisable at single regular frequencies, there were variations in wave form. The most commonly observed signal was biphasic and its wave form resembled the first time
100 F -4 -iC rC
80-
e
IL
n'
0
1
2
3 Time in Hours
6
5
4
Fig. 4 Change in electrical slow wave frequency with time in perfusion chamber. 11 AI
]A
II.I IIIII A%i/ui , !;1I 11I11I 1!I1111I111
1
1
iII,' I111111 111 11111 1 11
1
1I
II'III
- < C
I4I
lIt
III
Ii
v \~ ~ ~ ~ ~ ~I C
_ 1~~~~~~A/, u
In vitro electrical activity in canine colon' N. L. SHEARIN, K. L. BOWES,2 AND Y. J. KINGMA From the Surgical-Medical Research Institute and Departments of Surgery and Electrical Engineering, University of Alberta, Edmonton, Alberta, Canada
In vitro slow wave activity was studied in strips of right and left canine colon with silver/silver chloride electrodes. Using visual and computer analysis, slow wave frequency and coupling was assessed between different recording sites and the effect of a cholinergic agonist on coupling and frequency was determined. A regular slow wave was always found to be present. Frequency in the left colon was slightly higher than in the right with a slight decline noted with time. Spike activity was rarely seen in unstimulated specimens. Administration of a cholinergic agonist produced a decrease in frequency with no improvement in coupling. Coupling was usually better in a circular than in a longitudinal direction. It was concluded that if electrical activity is important in the control of colon contractions, it is more likely to be involved in the control of segmentation than in propagated contractions. SUMMARY
Electrical activity in the colon has been studied in vivo and in vitro in different experimental animals and in humans (Couturier et al., 1969; Weinbeck et al., 1972; Taylor et al., 1975; Snape et al., 1976). In most of these studies electrical slow waves have been recognised only intermittently at two or more frequencies. However, Christensen et al. (1969) found electrical slow wave activity in the in vitro cat colon to be omnipresent at 005 Hz (3 c/m). We have studied the in vitro electrical activity of muscle strips from canine colon. Coupling between different recording sites was assessed to determine whether electrical activity could be responsible for propagated contractions. Methods
Strips of colon (15 x 8 cm) were removed from healthy dogs anaesthetised with chloralose. The mucosa was removed by sharp dissection and the tissue placed in one of two tissue chambers (15 x *5 x 13 cm). A Krebs-Ringer solution kept at 370-37-50 and aerated with a gas mixture of 95% 02 and 5 % CO2 was allowed to flow through the tissue chambers and over the tissue at a flow rate of 5-6 ml/m. Eight glass insulated silver/silver-chloride 'Supported by the Medical Research Council of Canada. 2Address for reprint requests: Kenneth L. Bowes, MD, Department of Surgery, 11-1 17 Clinical Sciences Bldg, University of Alberta, Edmonton, Alberta, Canada, T6G 2G3. Received for publication 6 April 1979
electrodes (1 mm diameter) were gently pressed on to the circular muscle tissue surface until a satisfactory record was obtained. The electrodes were separated by a distance of 1I0 cm. A heavy silver wire was placed in the tissue chamber to serve as a reference electrode. This apparatus is similar to that previously described by Christensen et al. (1969). Recordings of electrical activity were made on a Beckman Dynograph R 411 polygraph (Beckman Instruments Inc., Shiller Park, Illinois, USA) with filters set to produce a pass band from 0 2 (12 cpm) to 30 Hz. The signals were stored on an SR 300 Ampex tape recorder (Ampex, Redwood City, Calif., USA). The low cutoff frequency is the result of the time constant which is introduced when the recorder is used in the a.c. coupled mode. This mode of coupling was required because of the large d.c. offsets which tend to fluctuate. On the equipment available the a.c. coupling introduced a time constant of one secondthat is, a corner frequency of 0 16 Hz. The roll off towards frequencies less than 0 16 Hz is not more than 6 dB/octave. Signals with a period of 0 05 Hz, for example, are attenuated to about one-third of their original value. A total of 54 strips of colon were studied. Thirtyfour strips were analysed in detail; 11 were cut in a circular direction (five right colon; six left colon) and 23 in a longitudinal direction (14 right colon; nine left colon). A 30 minute basal record was obtained from each specimen. The effect of neostigmine methylsulphate (0 3 Fm/ml) in the perfusion fluid was then determined. In four paired colon specimens (right and
780
781
In vitro electrical activity i'n cani'ne colon 0 0
r-I
0r~~~~
0
CC,3) v 0
0OC, .
0 0
0
0$8
0
0(O .
0
LO
0 0
0 0
CO) 0 0
CMJ
0
0
c,0
a
000 oo 0
0 0 e--i
K
0
:0 000co ILnne-)XXXXHHHnu
Fig.
0*0 00 000000 00
Typical frequency
(0-Cane
08$
Channel ;L
0l08888880-6-12-0-Bdooooooooooob'db-dooooooooooooooooooooOO00000-
power spectra
2;1
with the maximum values normalised
of the first channel having that value
of electrical activity recorded
to 1000 units.
is
left colon from four dogs) basal
was
If the values
in vitro
9
J0.
in two
from three
cycles/mmn.
or more
sites
sites
cm
apart
Vertical axis: arbitrary are
units
identical, only the symbol
printed.
recordings
continued for six hours. Electrical slow quency
orizontal axis;
Canel3)
determined manually by
wave
were
fre-
counting three
results
expressed as the number of specimens in same frequency was present in the first three recording electrodes (frequency comparison). are
which the
non-consecutive five minute segments of record from
The distance between the two outer electrodes
each electrode. A five minute interval
2
was
chosen
during periods of this length the frequencies significantly and because the interval was large enough for beat to beat variations to be averaged out. The records were also analysed on a digital computer. Before processing the data, the taped analogue signals from the instrumentation recorder were lowpass ifiltered (0-5 Hz cut-off frequency), digitised (sampling rate of one per second) and stored on magnetic tape. The digitised data were then processed on the digital computer to obtain the frequency power spectra (Jenkins and Watts, 1968). Dominant frequencies were obtained from these frequency power spectra. A typical frequency spectrum is shown in Fig. 1. Coupling between different recording sites was assessed by comparing frequencies and phase and determining coherence and cross-correlation. The first two assessments were made visually; the latter two by computer. 1. Frequencies obtained from different recording electrodes on the same specimen were compared. The because
did not appear to vary
was
cm.
2. The number of shift between two
cycles during which recording electrodes 2
the
phase
cm
apart
remained within 3600. 3. Coherence between
trodes, 2 quency
cm
apart
bands
was
(0-033
signals from pairs of eleccalculated for
Hz)
on
the
narrow
computer.
freCo-
herence is
a measure of the linear dependence of one signal on another. If one signal is completely dependent on the other, coherence is unity. Other-
wise the coherence may have values between
zero
and
unity. The mathematical definition of coherence (which holds for stationary processes only) is given by:
Where:
G,,,(f)
Gyy(f)
and
trum of the
G,,y(f)= one-sided the two
one-sided measured power spec-
signals. measured cross-power spectrum of
signals.
Coherence
was
determined in the frequency bands in
782
~ ~ m=
\.1 V... IF
1
n
N. L. Shearin, K. L. Bowes, and Y. J. Kingma
which the power spectra of both signals were maximal. The cross spectral density function G,y(f) is a complex function and can be expressed as:
1
lil
1'
IC.Y...
Gxy(f) = Cxy(f) +jOxy(f)
Where:
j=v-I Cxy(f)=co-spectral density function. Oxy(f)=quadrature spectral density function.
2
3
Fig. 2 Cross-correlation between pairs of channels. Horizontal axis: timeshift, I per division. Vertical axis: Arbitrary units. 1: good correlation. 2: fair correlation. OI
I1 11ittillll'lliiiili
I
III jII
11 IIIIIIIjJIIM II lti
ltil
lil III] III
IllMIli L
11 li til ll
11I 111
III11
IIIltl 11 I111
One may also write:
Gxy(f)= Gxy(f) ejOxy(f)
Where:
Gxy(f) = [CXY2(f) + OXY2(f)]
and:
Oxy(f)=arc tan Oxy(f) CXY(f)
In the case under consideration, Oxy(f) represents the phase angle as a function of frequency between the two signals (Bendat and Piersol, 1966). 4. Cross-correlation of slow wave activity measured at electrodes spaced 2 cm apart was calculated on the digital computer. Cross-correlation as a function of the time shift with X was displayed on a large scieen cathode ray tube and then photographed. Observations of the cross-correlation graph on the CRT enable classifying the results into one of three categories of coupling: good, fair, poor. An example is shown in Fig. 2. To facilitate comparison of I~I iI
iII4,Il
IIIltl1 11 III 1III I: '11 Illlitllls 111 IMHIIii111 L
I'ilIll
,ll
IIIIIII
I. {~~~~~~~~~~~~~ 1 min
Fig. 3 Electrical slow wave activity of the type most commonly observed in recordings from canine colon. Electrodes are arranged in a longitudinal orientation on the colon strips (Ch. 1-4= left colon; Ch. 5-7-= right colon).
In vitro electrical activity in canine colo7
783
results, a numerical grading system was used: (good=100%; fair=50%; poor=0%). The period (and thus frequency) and the phase shift between channels could also be obtained from the graphs. The cross-correlation function is defined by: +T (t) rnim 1 J fi(t)f2 (t-r)dr T->oo 2T -T Where f,(t) and f2(t) are the signals to be correlated. 5. Phase shift between different recording sites
was determined visually by using the relation
9=t x 3600, where t is the delay between the onset T of slow wave depolarisation at different recording sites and T is the period of a complete wave cycle. The phase angle was calculated only when coupling was present. Phase shift was calculated on the computer in those specimens in which the coherence value exceeded 0 5.
Results
A regular electrical slow wave was always found in strips of canine colon muscle studied in vitro (Fig. 3). Frequency in the left colon was only slightly higher than that in the right colon in both paired (4 90 ±0-63 and 4-13 +0-30 c/m) and unpaired (5-02 0-35 and 4-81 ±0-38 c/m) experiments. The differences were not significant (P >0.1). These frequencies are only slightly lower than those noted in the chronic in vivo dog colon (Bowes et al., 1978). The in vitro frequency was well maintained in the tissue chamber. Only a slight decline was noted with time (Fig. 4). Spike activity was rarely seen in unstimulated specimens. Although slow waves were always recognisable at single regular frequencies, there were variations in wave form. The most commonly observed signal was biphasic and its wave form resembled the first time
100 F -4 -iC rC
80-
e
IL
n'
0
1
2
3 Time in Hours
6
5
4
Fig. 4 Change in electrical slow wave frequency with time in perfusion chamber. 11 AI
]A
II.I IIIII A%i/ui , !;1I 11I11I 1!I1111I111
1
1
iII,' I111111 111 11111 1 11
1
1I
II'III
- < C
I4I
lIt
III
Ii
v \~ ~ ~ ~ ~ ~I C
_ 1~~~~~~A/, u
